it continues to be the most compelling analogy for me — federation of task specific models and classical systems to just pixels in policy out
MACHINE LEARNING
-
AI revolutionizes science, creativity, and discovery, not just cats.
By
–
Sorry, but with everything happening in AI, how can people still think that AI is 'just pictures of cats talking'? AI is completely redefining the possible boundaries of science, creativity, ingenuity, and discovery.
-
Error Recovery Rate Gap in AI Task Performance
By
–
Passed and failed tasks encounter similar errors (2.09 vs 2.71 per task). The difference? • Passed tasks recover from 95.0% of errors • Failed tasks recover from 73.5% A 21.5-point gap. Recovery > avoidance.
-
SWE-Rebench: Software Engineering AI Evaluation Benchmark January 2026
By
–
The eval is probably not perfect, so please just consider it as an additional signal https://
swe-rebench.com/?insight=jan_2
026
… -
Chinese LLMs Performance on SWE-bench: M2.5 Underperforms Opus
By
–
Oof, SWE-rebench is brutal for recent Chinese releases M2.5 reported 80.2% on SWE-bench verified against 80.8% for Opus 4.6, but it seriously underperforms here Qwen3-Coder-Next looks good with 40% and only 80B A3B parameters
-
OpenSearch 3.4 Release: Advanced Search and AI Capabilities
By
–
Introducing OpenSearch 3.4 https://
bit.ly/4jth9zv
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
GPT-5.2 Discovers New Gluon Interaction in Theoretical Physics
By
–
GPT-5.2 derived a new result in theoretical physics. We’re releasing the result in a preprint with researchers from @the_IAS
, @VanderbiltU
, @Cambridge_Uni
, and @Harvard
. It shows that a gluon interaction many physicists expected would not occur can arise under specific -
Bahdanau Attention: The Breakthrough Behind Transformer Architecture
By
–
Bahdanau, Cho and Bengio, authors of the original attention paper. This is the breakthrough. The transformer paper is tweaks on it.
-
2026: AI Acceleration Begins With Gemini 3 Deep Think Breakthrough
By
–
2026 est une année historique : c’est l’année du début du bouleversement civilisationnel lié à l’IA On attendait un ralentissement de l’IA À la place on a une accélération fulgurante des progrès de l’IA Les performances en codage de Gemini 3 Deep Think sont stupéfiants : il
-
Codex optimization nerfs ML predictive accuracy
By
–
Quick update: seems like Codex tried to “optimize” the ML library by *massively* nerfing predictive accuracy. Trying to fix that now.